User:Giorgio italy/AutoDispatcher/The model

The model

Full automation of a layout requires that the computer be able to:

   Operate trains, turnouts and signals (if installed).

   See trains, i.e. track their position as they move through the layout.

The first task is made easy by installing DCC decoders and connecting the command station to the computer through an appropriate interface.

Turnouts and signals can be alternatively controlled by other systems, like C/MRI or CTI, which already include a computer interface.

The second point requires installing sensors, which come in different flavors and can be grouped into two main families:

   Occupancy sensors
   Based on current detection circuits, they sense the presence of a train (or at least its locomotive) on a track segment of any length.

   Transit sensors
   Implemented using photocells or reed/mechanical switches, they detect the transit of a train though a specific spot.

Any of the above can be used and even mixed, although occupancy sensors offer some advantages and appear to be better supported by JMRI.

Whatever type of sensors is chosen, a suitable feedback bus must connect them to the command station, or directly to the computer.

How many sensors are needed?

The minimum number of sensors per block depends on whether trains are run by the Engineer component of AutoDispatcher or whether they are controlled manually or using a custom script:

   The Dispatcher component requires at least one sensor per block (two if transit sensors are used).

   The Engineer component requires at least two sensors per block (irrelevant whether occupation or transit sensors are used).

The second sensor required by the Engineer component is used to halt the train exactly in front of block's exit signal. This is not very different from what happens in the prototype, where an additional sensor is used to stop the train at the right place: the eyes of the engineer :-)